This Semiconductor Stock Could Be a Better Way to Invest in AI and Satellites Than SpaceX

STMicroelectronics is positioned to benefit from SpaceX's planned AI-compute and Starlink expansion: its data-center business is projected to generate $1 billion of revenue in 2026 and more than double in 2027, while SpaceX-related revenue is expected to total $3 billion during 2026-2028. SpaceX targets roughly 10GW of compute capacity by end-2027 and could launch about 100,000 additional Starlink V3 satellites over coming years, expanding demand for STM's optical-connectivity, RF, power-conversion and thermal-management chips. STM has guided to $18 billion in revenue by 2028, and the article argues its EPS could nearly triple through 2028, making its roughly 37x forward P/E more attractive than SpaceX's estimated 100x sales valuation.

Analysis

The central underwriting issue for STM is not whether AI/LEO spending grows, but whether it converts into a sufficiently concentrated, high-margin revenue stream to offset its much larger auto and industrial exposure. A $1B data-center business would be material, but it is unlikely to command AI-infrastructure multiples unless STM discloses design wins, content-per-server, and gross-margin characteristics that distinguish it from lower-value power and connectivity peers. The market will likely discount unverified customer-specific projections until backlog, utilization, and segment margin data appear in results.

Near term, the stock remains more sensitive to the semiconductor inventory cycle, European auto production, China demand, and SiC pricing than to a multi-year space narrative. The non-obvious downside is that hyperscaler capex can favor integrated optical and networking vendors such as AVGO, MRVL, MTSI, LITE and COHR, while satellite volumes may face launch cadence, qualification, and radiation-hardening bottlenecks; unit growth does not automatically equal STM revenue growth. Conversely, credible disclosure that data-center revenue is growing faster than planned while fab utilization recovers could drive both EPS revisions and multiple expansion over 6-18 months.

Consensus appears too willing to annualize a prospective customer relationship and too dismissive of cyclicality. STM at a premium forward multiple needs evidence that incremental revenue is structurally accretive; absent that, an auto/industrial downturn can compress both earnings and valuation before the AI/space opportunity is realized. The key falsifiers are a cut to 2027-28 revenue targets, flat-to-down gross margin despite utilization improvement, or customer concentration/backlog commentary that shows the opportunity is not scalable beyond one program.

Key Decisions for Investors

• Do not chase STM on promotional AI/space headlines. Establish a 6-12 month watch position only after the next earnings release confirms data-center revenue, order visibility, and gross-margin contribution; scale if management raises medium-term targets rather than merely reiterating them.
• For a cleaner AI-connectivity expression over 3-9 months, favor a basket long AVGO/MRVL/MTSI versus STM until STM provides independently verifiable optical design-win and revenue disclosures. Review the spread after each company reports AI-networking backlog and gross-margin trends.
• Use STM downside as an auto/industrial-cycle hedge: if European auto production indicators weaken or STM guides utilization lower, pair short STM against long higher-quality AI infrastructure exposure such as AVGO. Cover if STM reports sequential utilization recovery plus gross-margin expansion.
• Set an event alert for disclosed SpaceX-related backlog, customer concentration, and 2027 segment revenue guidance. A documented multi-year backlog with margins above corporate average would invalidate the cautious stance and justify upgrading STM from watch position to core long.